CASE STUDY / 01
MULTI-STRATEGY QUANTITATIVE TRADING ENGINE
APEX QUANT ENGINE
Multi-strategy quantitative trading engine.
Apex is a multi-strategy quantitative trading engine spanning BTC, ETH and SOL. It couples a market-data engine, a strategy engine, a risk engine, portfolio allocation and deterministic backtesting with paper execution — 11.9M+ one-minute bars and 5,867 completed backtest trades.
EVIDENCE INSTRUMENT / PROJECT-SPECIFIC
INSTRUMENT / EQUITY SURFACE
ILLUSTRATIVE — SEEDED DETERMINISTIC SHAPE, NOT THE BACKTEST ARTIFACT
The surface demonstrates the shape of a multi-strategy equity trajectory with its drawdown topology. Hover to inspect any point. The measured master-fund result appears in the results chapter below — a backtest outcome at a single point in history, not a forecast.
PROBLEM
Retail-grade backtesting tools make it easy to produce numbers that cannot be reproduced. Lookahead bias, implicit assumptions about fills and fees, and non-deterministic data access quietly inflate results.
Worse, most frameworks couple strategy logic to execution logic, so the code that generated a backtest is not the code that would trade — the simulation and the system drift apart.
MOTIVATION
I wanted to know whether a systematic, multi-strategy approach to crypto markets could survive honest treatment: exact costs, deterministic replay, and risk caps enforced before position sizing rather than after losses.
The engine exists to answer that question with evidence rather than narrative.
RESEARCH QUESTION
Can a multi-strategy crypto portfolio, simulated over millions of one-minute bars with explicit fees, slippage and hard risk caps, remain net-positive after all modeled costs?
OBJECTIVE
- Build a single engine where research, simulation and (paper) execution share one deterministic code path.
- Enforce risk at the portfolio level before any order is constructed.
- Make every backtest reproducible from raw data to final equity curve.
ARCHITECTURE
VISUAL METAPHOR — PORTFOLIO / STRATEGY / EQUITY LANDSCAPE
ARCHITECTURE — DATA PATH
ARCHITECTURE DIAGRAM — THE DOCUMENTED LAYERS, NOT A MEASUREMENT
- 01
DATA ENGINE
Market-data ingestion and normalization
- 11.9M+ 1-minute bars across BTC / ETH / SOL
- Deterministic, replayable access layer
- 02
STRATEGY ENGINE
Multi-strategy signal generation
- Independent strategies evaluated per asset and timeframe
- Signal contracts decoupled from execution
- 03
RISK ENGINE
Pre-trade risk enforcement
- Position sizing derived from risk budget
- Hard caps applied before order construction
- 04
PORTFOLIO ALLOCATION
Cross-asset capital distribution
- Allocation across strategies and assets
- Master-fund composition for backtests
- 05
EXECUTION
Deterministic fills and paper execution
- Fee and slippage modelling at fill time
- Paper execution mode for forward testing
- 06
BACKTEST CORE
Deterministic simulation harness
- Reproducible run-to-run results
- 44.50s execution benchmark for full master-fund backtest
DATA FLOW
PIPELINE — 6 STAGES, IN ORDER
- 01Historical 1-minute bars are loaded through the deterministic data engine
- 02The strategy engine evaluates every strategy per asset and timeframe, emitting signal contracts
- 03The risk engine converts signals into position sizes bounded by the risk budget
- 04Portfolio allocation distributes capital across the strategy set
- 05The execution layer simulates fills with explicit fee and slippage models
- 06The backtest core aggregates fills into trades, equity and drawdown series
METHODS
- Deterministic simulation — identical inputs and code produce identical equity curves
- Explicit cost modelling — fees and slippage charged at every simulated fill
- Multi-strategy composition across BTC / ETH / SOL with portfolio-level allocation
- Paper execution for forward validation of the same code path used in backtests
RISK / VALIDATION
- Risk budgets are enforced before order construction, not reconciled after losses
- Position size derives from stop distance and per-trade risk cap
- Every backtest result is traceable to raw bars, parameters and code revision
EXPERIMENTS
Master-fund backtest
COMPLETEDWhat is the net simulated ROI of the full multi-strategy portfolio after all modeled costs?
+570.18% net simulated ROI across the master-fund backtest.
Full-history simulation
COMPLETEDHow does the engine behave across the complete available 1-minute-bar history?
5,867 completed trades over 11.9M+ bars; full backtest executes in 44.50 seconds.
Paper execution
ONGOINGDoes the same code path trade consistently in forward paper mode?
Paper execution mode operational for forward validation.
RESULTS
+570.18%
NET SIMULATED ROI
BACKTEST
MASTER-FUND BACKTEST
Self-reported master-fund backtest after modeled fees and slippage; evaluation window and seed not documented
5,867
COMPLETED BACKTEST TRADES
BACKTEST
FULL-HISTORY SIMULATION
11.9M+
1-MINUTE BARS PROCESSED
BACKTEST
BTC / ETH / SOL
44.50s
BACKTEST EXECUTION TIME
SELF-BENCHMARK
MASTER-FUND RUN
Self-reported single-run benchmark; machine specification not documented
All figures are simulated. The +570.18% figure is the net result of the master-fund backtest after modeled fees and slippage — it is not a live trading return and should not be read as one.
The engine's value is less the number than the discipline: deterministic replay, explicit costs, and risk enforced before sizing.
LIMITATIONS
- All results are backtests and paper simulations — no live trading has occurred
- Cost models approximate real fills; extreme liquidity conditions are not fully captured by fee/slippage models
- Historical performance of a strategy set does not imply future performance
CURRENT STATUS
BACKTEST · PAPER
Deterministic backtesting and paper execution. Not live trading.
EVIDENCE
BACKTEST ARTIFACTS
Equity curves, trade logs and run configurations reproducible from the repository
OPEN THE SOURCE →PAPER MODE
Forward paper-execution mode runs the identical code path
INTERACTIVE LAB
EQUITY CURVE + DRAWDOWN — SEEDED ILLUSTRATION OF THE MASTER-FUND BACKTEST SHAPE
R-MULTIPLE DISTRIBUTION — SEEDED ILLUSTRATION OF THE TRADE POPULATION SHAPE